Divvy is an application for performing unsupervised machine learning and visualization. We focus on the clustering (separating data into groups) and dimensionality reduction (finding low dimensional structure in high dimensional data) subfields of machine learning. For visualization we provide support for both the whole dataset (e.g. a scatter plot) and points (e.g. transforming a particular point into an image).

* Have lots of cores? Yeah, we'll use those.

Divvy is both task and data parallel. No longer will you be waiting for one algorithm to complete before you start another. Start as many as you want and keep using the UI. Only started one? With data parallelism we'll still push your new MacBook Pro to 800% CPU utilization.

* Part of your workflow.

Export your clusterings and reductions to .csv and your visualizations to .png. Use your Matlab or R data with our Matlab/R to Divvy export tools available at http://github.com/jmlewis/divvy.

For more info find us on the web at http://divvy.ucsd.edu.